Wednesday, 17 May 2017

Etching Tank

I've been busy on non-ham activities the last few weeks so I haven't progressed many projects. I'm re-organising the workshop and trying to throw out some equipment surplus to requirements. My wife calls it junk!

I had a few SCSI tape drives lying around and as I opened one up to recover the controller board it occurred to me I could make a tank agitator for etching. In the past I had used an aquarium bubbler with an air stone. I stopped using this because it generated a plume of etchant that stained nearby objects. I moved on to putting the pcb in a zip lock back and manually rocking it while it sat in a dish.

This works well but if I could automate the rocking and heat it at the same time things would be better. I'm blessed with a working stove and hot plates in the workshop, but don't tell the wife or I might be expected to move into the workshop for good.

So there when I opened the SCSI drive was a great big cog used to load the tape. It took all of 5 minutes to work out there was a lug underneath the cog I had to chop off so it could rotate continuously and wire up a plug pack to the motor. A few more minutes to work out how to attach the rotating cog to the etchant bath and I was done. I used a bolt onto a pop stick, or small piece of wood found in frozen ice sweets, and a piece of string from the wood to a peg on the dish.

With the hotplate turned on and the motor running things went smoothly. The only issue is the dish tends to walk around the surface of the hot plate. I'm going to fix that by running the cord over an arm and pulley perhaps 50cm above the etchant tank. That way the lifting on the tank will be vertical instead of including a side to side motion as well.

The current arrangement looks like this before I make the arm and pulley:

A very temporary proof of concept. Which just goes to show that one man's junk can contain his treasure.

Regards
Richard VK6TT

Friday, 5 May 2017

Heatsink Fan Temperature Switch

As I was pondering the next step with my bag of 2SC5707 transistors the problem of heat-sinking occurred to me. An elegant solution would be to mount the transistors to the PCB but the cooling is awkward. Then I remembered how effective forced air cooling is and remembered I hadn't shared this with readers before.

 I put most commercial class VHF FM transceivers way ahead of ham gear when it comes to selectivity and immunity from pagers. The only drawback I have ever noticed was with regard to transmitting duty cycle. Generally, the commercial transceiver was never intended for long overs and the heat-sinking reflects this. Fine for repeater use but if you have a long over on simplex things get hot. Sometimes very hot!

Here is a simple project that really protects that radio. First, the results of before and after measurement of heat-sink temperatures are shown below on a Philips PRM80 transmitting 25 watts into a dummy load:


It is true that many transmissions do not last 3 minutes or longer. However, a high SWR at the radio will increase the heat dissipated by the final transistor. I generally only check my antenna when I suspect it needs to be checked. So I could be transmitting for some time before any problem is detected. And while I have never sat on the microphone I am sure we have all heard someone driving around with the transmitter keyed up for extended periods.

I consider this a worthwhile addition to any PMR type radio. From the circuit below you will see that when the output of the temperature sensor reaches a set point the fan is turned on. Once the temperature falls the fan turns off.

I made mine on a small 0.8mm thick piece of PCB so that the surface mount LM61 could be mounted in a slot in the board to make contact with the heat-sink when the pcb was bolted to the heat-sink.

I'm going to use the same circuit on a forthcoming experimental HF amplifier using all of my remaining 2SC5707's. One last hurrah for them or a successful 15 watt linear amplifier for the lower HF bands.

Regards
Richard VK6TT

Friday, 28 April 2017

Test Equipment Tip for Initial Testing of Transmitter Amplifiers

While many of us start out building transmitter stages and tuning for maximum output, we should quickly learn that over-driving a transmitter stage must be avoided. Even a slightly over driven amplifier stage generates significant harmonics and distortion products. So here is a quick tip on using the oscilloscope to check for the presence of a transmitter stage being over-driven. 

A 10MHz analogue oscilloscope will be fine for looking at HF signals since we are going to measure relative voltages. I am unsure if a low bandwidth digital CRO will work.

Connect your oscilloscope to the collector or drain of the active device. With no drive and the oscilloscope set to AC coupling ensure the horizontal trace is on the centre graticule. The time-base setting is not critical so I use something that gives me a band across the screen with drive applied rather than the actual waveform. Try 1ms per division and adjust to suit your preference.

Now apply perhaps 10% of the drive you expect to use when the amplifier is in operation. Adjust the trace so it sits between say the second graticule above the centre line and the second graticule below the centre line. As you increase the drive you should see the trace touch the third graticule above and below the centre line at the same time.

Adjust the vertical amplifier gain to reduce the signal size on the screen then increases the drive again. Again, you should see the trace touch the third graticule above and below the centre line at the same time. Keep repeating this until you notice that one of the third graticule's above or below the centre is being touched before the other is being touched.

At this point the amplifier is no longer linear. If you have a fast oscilloscope you can increase the time-base speed to observe a few cycles of the waveform. You will notice that when non-linear the waveform peaks that were last to hit the third graticule will be distorted.

You can back off the drive until both the positive and negative peaks are moving in unison, which I find easier to discern with a slow timebase setting rather than a few cycles being displayed on the screen. That is the limit for linear operation of the stage.

Spectrum analysers and other nice test gear allow you to measure how non-linear your transmitter is. But isn't it easier just to avoid over-driving the transmitter to begin with? Hopefully this tip helps you avoid over-driving your transmitter stage giving you a nice clean signal when you finally go on air.

Regards
Richard VK6TT

Tuesday, 25 April 2017

250mW Class A Amplifier with 2SC5707

Further to my recent posts on using the 2SC5707 of unknown origin I finally got around to some more testing. I calculated the component values for a 250mW Class A amp and after etching a circuit board built the amplifier for testing. Overall, the results were pleasing and justify further experimentation.

Referring to my bible, "Solid State Design" by Hayward and DeMaw, I expected a gain of 17dB from the circuit below:



I achieved this 17dB of gain at 10MHz. However, the gain had fallen by 4dB at 20MHz so at present I will limit my use of this transistor to 14MHz and lower.

My clip on heatsink  was a 15mm x 15mm piece of circuit board, held on with a wooden clothes peg. Worked extremely well for testing but I have a more permanent solution in mind which I will post about shortly.

Overall a nice little amplifier and I will box it up for use on the workbench to increase the drive from my signal generator as required. But ironical that it will be a 15c transistor between two expensive BNC connectors.

With 8 transistors left I will really push them in the next experiment to see if there is any smoke in them!

Thursday, 30 March 2017

Testing my Noise Source and 2m Band Pass Filter - update

Yesterday I caught up with Jack, VK6KDX's, and enjoy the pleasure of chatting about his projects while drinking his coffee. A great way to spend a few hours.

One of the activities we pursued was looking at the output spectrum of noise sources on some test gear Jack owns. Jack demonstrated his noise source and it works very well for testing 23cm filters. We then had a look at my noise source on the spectrum analyser. It had peaks and looked horrible. It occurred to me this morning that the reason for that was most likely stray RF being coupled into the noise source and being amplified by the mmic chain. I still haven't put my noise source into a metal box and the power supply decoupling could be greatly improved.

Despite the shortcomings of the test set-up it did allow me to see the 2m Band Pass Filter in action. A small shoulder on the left of the response curve could be either stray RF coupling or a result of this being a no-tune filter. But I'm really happy with both the noise source and the filter. Especially the filter. It's always pleasing when you build something that confirms your measurement technique for the inductors is really good. No trim-caps, no squeezing or expanding coils. Measure, design and build. Worked first time with no tweaking.






The spectrum analyser was set to 146MHz, 20Mhz per division. We didn't check or adjust for any drift in the spectrum analyser itself. As expected the response falls off faster on the high side. If you haven't read how I measure nH inductors then check out the post http://vk6tt.blogspot.com/2017/01/measuring-small-inductors.html.

Regards
Richard VK6TT

Thursday, 23 March 2017

Testing my Noise Source and 2m Band Pass Filter

Well it turns out I couldn't readily find anyone with the test gear to look at my homebrew noise source. I will keep asking around but in the meantime I thought I'd see if it could be used to determine the response of the 2m bandpass filter I described in this post.

The first thing I did was look at the noise source output across the spectrum in 10MHz bites with my spectrum analyser. It wasn't flat. The was about +/- 2dB from 10MHz to 1GHz.  So any measurements would have to be on a before and after basis.

So I put the 2m filter between the noise source and the spectrum analyser and used the peak hold function as a surrogate for an average over time.  I measured the response every 10MHz from 85MHz to 205MHz. Then I  removed the filter and measured the levels again. Subtracted the difference and I had a response curve.

Here then is the measured results against what was modelled:

If you look closely the measured response is indicated by the 'dot' at each marker. Clearly something is not quite right. Somehow the filter has "gain" at 135MHz and 145MHz. That somehow is not possible. And the measurement at 125MHz looks odd.

I had previously measured the filter response into a good 50 ohm power meter and it was a very close match to what was modelled. Hence, the distortion in the measured results was due to either the spectrum analyser not being a true 50ohm resistive input or the patch leads were not 50 ohms. Either way it shows that the noise source can be used for measurement, but only when the test gear is not influencing the result.

Perhaps the reason for the +/- 2dB response a cross the spectrum is due to the spectrum analyser input impedance. Once I find another piece of test gear to measure the noise source on I will know more. In the meantime, I'm using the mythbusters approach and calling the noise source project finished.

Since this noise source costs peanuts I urge you to build one, or something similar, and comment on how it went.

Regards
Richard VK6TT




Friday, 17 March 2017

Spotting Fake Tansitors from China - update

It had been just on 4 weeks since I ordered the suspect transistors from China, as described in the entry http://vk6tt.blogspot.com/2017/01/spotting-fake-transistors-from-china.html.

To recap, the 2SC5707 might be suitable for QRP HF transmitters up to perhaps 1 watt. I discussed on the blog referred to above why none of the 81 listings for this transistor on Aliexpress appeared genuine. Which is a pity, because the datasheet for the transistor suggested it had promise.

In the quest to uncover the truth, or confirm my suspicions, I purchased 10 pieces from the seller with the highest rating to see what transpires. Today was the first chance I have had to pursue this since those transistors arrived. A cursory examination showed they were different from what was in the listing photo. The front face did not have the dimple the listing showed, while the reverse side had a different heatsink.

Undeterred, I measured the hfe for the 10 transistors and found it to be between 230 and 270. I quickly substituted one into a broadband utility amplifier I was using to test a push pull amplifier for 3.5MHz. I never expected much from the transistor and therefore the equipment currently on the workbench was pressed into service. With this equipment the minimum frequency I could measure gain at  was 10MHz. Since the final transistor on this utility amplifier was never going to perform well above 3.5MHz I simply did a few quick before and after measurements to see if a more thorough test was worthwhile.

The utility amplifier is a class A design with three stages. A mmic drives the 2N2219A which drives a 2SC3230. According to the datasheet the 2SC3230 is intended to be a final stage for cordless phones at 1.7MHz and has a ft of 100MHz.

What I found surprised me. After a few reference level checks the 2n2219A transistor was removed and replaced with the 2SC5707. The smoke stayed inside the transistor and there was RF coming out of the amplifier at almost the same level as before.

 Compared with the original transistor I found the gain changed as follows:



Change in output level  of utility amplifier after 39dB of attenuation to output level
Frequency
2N2219A (Original)
2SC5707
Difference
10MHz
-17 dBm
-19 dBm
-2 dB
20MHz
-27 dBm
-29 dBm
-2 dB
30MHz
-35 dBm
-39 dBm
-4 dB

So it looks like these transistors, which cost cents, might have potential. My intended application is as the final stage in a class C amplifier for a WSPR beacon on 28MHz. This quick test has not ruled them out for this application. I will put together a test amplifier for class C at 28MHz and report later on my findings.  

By way of comparison the cost when buying 10 pieces of a 2N2219 is around 60c each. The 2SC2314 works out at around 40c and the 2SC5707 is under 20c each.




Regards
Richard VK6TT